HackerShelf
151 books Hacker News keeps recommending, sifted from 9.4 million comments: a static site Cursor built in two and a half hours.
HackerShelf sifts 151 books Hacker News keeps recommending out of 9.4 million comments. Its author, IshankDev, created the repo on November 2, 2026, and wrote up the build on November 4: using Cursor, the whole thing took two and a half hours, with the data pipeline, the site, and the deploy done in about 40 minutes. The GitHub repo just launched and has no stars yet. The filter follows Karpathy's HackerTube rule: a book only counts if at least three different people mentioned it in at least three different months. The raw comments run from January 2024 to August 2026 — 9,437,609 of them — from which Gemini pulled 6,076 book mentions in 288 batches; Open Library matched 2,212 works, leaving 151 books in 9 categories. The site itself is a static page you just open and use: search by title, author, or topic, and every book links back to the Hacker News comment that mentioned it; the author spot-checked 83 quotes across 10 books against the original comments, and all 83 checked out. The top book is Designing Data-Intensive Applications, recommended by 13 people and mentioned 14 times, and you can save books to a reading list stored locally. Two caveats: only comments from 2024 onward were scanned, so recommendations from before 2024 aren't counted; and the site is static with no backend account, so your reading list lives only in your own browser.
